您的位置：首页 > 论文页面
一种快速井位优化的代理模型
发表时间：2019-10-31 浏览量：1303 下载量：145
| 全部作者： | 唐力辉，李俊键，王依诚，陆文明，姜汉桥， 廉培庆，苏海婷，周宇 |
| 作者单位： | 中国石油大学（北京）石油工程学院，油气资源与探测国家重点实验室；中国石化石油勘探开发研究院 |
| 摘 要： | 基于优化理论的井位优化设计需要运行大量的布井方案，尤其是对于大规模的模型，减少数模时间、提高计算速度至关重要。针对优化迭代过程计算时间太长的问题，本文采用代理模型来解决，提出了两种建立代理模型的方法，第一种方法是结合油藏工程知识，利用递减曲线和水驱曲线拟合，预测累计产油量、累计产水量，进而预测目标函数值，如净现值（net present value，NPV）；第二种方法是利用三次样条插值方法建立的代理模型，直接对目标函数进行插值预测。通过典型的测试模型对这两种方法进行了准确性和有效性的验证，结果表明，三次样条插值预测的效果更好，并对该代理模型的误差容忍度进行了对比，确认误差容忍度在5%时就能保证预测的准确性，也能显著减少数模运行时间。 |
| 关 键 词： | 能源科学技术其他学科；井位优化；代理模型；曲线拟合；三次样条插值；容忍度 |
| Title： | An agent model for fast well location optimization |
| Author： | TANG Lihui, LI Junjian, WANG Yicheng, LU Wenming, JIANG Hanqiao, LIAN Peiqing, SU Haiting, ZHOU Yu |
| Organization： | State Key Laboratory of Petroleum Resources and Prospecting, College of Petroleum Engineering, China University of Petroleum (Beijing)；SINOPEC Petroleum Exploration and Production Research Institute |
| Abstract： | Well location optimization design based on optimization theory will run a large number of well placement schemes. How to reduce the digital-to-analog time and increase the calculation speed is fatefully important, especially for large-scale models. In order to solve the problem that the optimization iteration process is too long, the proxy model is used in this paper. Two methods for establishing the proxy model are proposed. The first method is to combine the reservoir engineering knowledge and to use the decreasing curve and the water drive curve to fit, predicting the oil production and the water production, and then to predict the objective function value such as net present value (NPV). The second method is to use the proxy model established by the cubic spline interpolation method, to directly interpolate the objective function. The accuracy and validity of the two methods are verified through a typical test model. The results show that the cubic spline interpolation prediction is better, and the error tolerance of the proxy model is compared. It is confirmed that when the error tolerance is guaranteed at 5%, the accuracy of the prediction can be ensured and the digital-to-analog runtime can be significantly reduced. |
| Key words： | other subjects of energy science and technology; well location optimization; proxy model; curve fitting; cubic spline interpolation; tolerance |
| 发表期数： | 2019年10月第5期 |
| 引用格式： | 唐力辉，李俊键，王依诚，等. 一种快速井位优化的代理模型[J]. 中国科技论文在线精品论文，2019，12（5）：837-842. |
6
评论数 0请您登录
暂无评论
下载全文
多维论文